Crate langextract_rust 

LangExtract

A Rust library for extracting structured and grounded information from text using LLMs.

This library provides a clean, async API for working with various language model providers to extract structured data from unstructured text.

Features

• Support for multiple LLM providers (Gemini, OpenAI, Ollama)
• Async/await API for concurrent processing
• Schema-driven extraction with validation
• Text chunking and tokenization
• Flexible output formats (JSON, YAML)
• Built-in visualization and progress tracking

Quick Start

use langextract_rust::{extract, ExampleData, Extraction, FormatType};

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let examples = vec![
        ExampleData {
            text: "John Doe is 30 years old".to_string(),
            extractions: vec![
                Extraction::new("person".to_string(), "John Doe".to_string()),
                Extraction::new("age".to_string(), "30".to_string()),
            ],
        }
    ];

    let result = extract(
        "Alice Smith is 25 years old and works as a doctor",
        Some("Extract person names and ages from the text"),
        &examples,
        Default::default(),
    ).await?;

    println!("{:?}", result);
    Ok(())
}

Re-exports

pub use config::LangExtractConfig;

pub use config::ProcessingConfig;

pub use config::ValidationConfig as NewValidationConfig;

pub use config::ChunkingConfig;

pub use config::AlignmentConfig as NewAlignmentConfig;

pub use config::MultiPassConfig as NewMultiPassConfig;

pub use config::VisualizationConfig;

pub use config::InferenceConfig as NewInferenceConfig;

pub use config::ProgressConfig;

pub use config::ChunkingStrategy;

pub use config::ExportFormat as NewExportFormat;

pub use data::AlignmentStatus;

pub use data::AnnotatedDocument;

pub use data::CharInterval;

pub use data::Document;

pub use data::ExampleData;

pub use data::Extraction;

pub use data::FormatType;

pub use exceptions::LangExtractError;

pub use exceptions::LangExtractResult;

pub use inference::BaseLanguageModel;

pub use inference::ScoredOutput;

pub use logging::ProgressHandler;

pub use logging::ProgressEvent;

pub use logging::ConsoleProgressHandler;

pub use logging::SilentProgressHandler;

pub use logging::LogProgressHandler;

pub use providers::ProviderConfig;

pub use providers::ProviderType;

pub use providers::UniversalProvider;

pub use resolver::ValidationConfig;

pub use resolver::ValidationResult;

pub use resolver::ValidationError;

pub use resolver::ValidationWarning;

pub use resolver::CoercionSummary;

pub use resolver::CoercionDetail;

pub use resolver::CoercionTargetType;

pub use visualization::ExportFormat;

pub use visualization::ExportConfig;

pub use visualization::export_document;

pub use pipeline::PipelineConfig;

pub use pipeline::PipelineStep;

pub use pipeline::PipelineResult;

pub use pipeline::PipelineExecutor;

Modules

alignment

Text alignment functionality for mapping extractions to source text positions.

annotation

Text annotation functionality.

chunking

Text chunking functionality for processing large documents.

config

Unified configuration system for LangExtract.

data

Core data types for the annotation pipeline.

exceptions

Error types and result definitions for LangExtract.

factory

Factory for creating language model instances.

inference

Language model inference abstractions and implementations.

io

I/O utilities for loading text from various sources.

logging

Logging and progress reporting system for LangExtract.

multipass

Multi-pass extraction system for improved recall and quality.

pipeline

Pipeline processing for multi-step information extraction.

progress

Progress tracking functionality.

prompting

Advanced prompt template system with dynamic variables and provider adaptation.

providers

Language model provider implementations.

resolver

Output resolution and parsing functionality.

schema

Schema definitions and abstractions for structured prompt outputs.

templates

Template engine and utilities for LangExtract.

tokenizer

Text tokenization functionality.

visualization

Visualization utilities for annotated documents.

Macros

progress_debug

progress_error

progress_info

Convenience macros for common progress events

Structs

ExtractConfig

Configuration for the extract function

Functions

extract

Main extraction function that mirrors the Python API

extract_with_config

Convenient extraction function using the new unified configuration

visualize

Visualize function that mirrors the Python API